Artificial intelligence is a threat to jobs or it has the power to change the world for the better, or – depending on which report you read – both.

While the rise of AI has been the subject of extensive stories in the popular media, what frequently goes unmentioned is the crucial component that drives these intelligent machines – data.

As a result, the role data plays in enabling deep learning to function – and thus to shape business decisions – is largely ignored.

But we are living in an age where there is an abundance of readily available data that can be easily interpreted, and which exists both inside and outside of an organization’s four walls.

Once businesses grasp that, it becomes easier for them to understand how they can use automation to provide a better service to their customers.

However, understanding the significance of your available data and putting it to good use are two different things. And, before we can harness the power of data, we must prepare ourselves at an organizational level to get the most from it.

The role data plays in enabling deep learning to function – and thus to shape business decisions – is largely ignored.

Deep dive into an ocean of data

We have only recently been able to properly mine data by using “deep learning” – a type of AI where machines begin to understand and recognize patterns rather than follow a set of ordered tasks.

This has made it easier to interpret quickly a variety of data, such as flat, textual or image-based information, that has overtaken many once state-of-the-art data science and statistics software products.

These have been replaced by a proliferation of open-source deep-learning tools, such as data visualization – the ability to extract meaningful information from unstructured or badly structured sources of information – that digitally agile companies can use to their benefit.

But to make the most of data’s possibilities, business leaders must be able to use their imaginations to come up with a range of data-driven “cyber opportunities” while being aware of possibilities and limitations of data science.

A big challenge here is balancing business expectations, such as extracting value from the existing company during its transformation to the data-driven economy, while measuring and delivering improved, and perhaps different, services.

Here, it’s good to mention that internal data about your clients can be just as valuable, if not more so, than often-alluring public information about wider communities. The key is to find new and effective ways to tap into your existing data sources.


Don’t forget about data science

As companies continue to develop their digital transformation strategies to compete in the Business 4.0 landscape, many are forgetting the primary importance of data science.

The appeal of potentially game-changing digital technologies is obvious. But it’s a mistake to adopt these technologies before laying foundations that are solidly based on reliable data.

A company that ploughs ahead with an AI initiative based on machine data without the organizational maturity to formulate, implement, go live with – and continuously improve – its services is doomed to failure.

Companies that adopt this approach are essentially looking through the wrong end of the telescope.

The appeal of potentially game-changing digital technologies is obvious.

Adopt a business-first mindset

To look through the right end of the telescope, you need to be focused on the business problem you want to solve, rather than the technology you want to use: you must “think past the problem”.

This means working out what your desired outcome will be: is the goal to process automation for efficiency, or to amplify human effectiveness with contextual insights? Or is the digitalization of activities something that requires computational as well as predictive power, for example in real-time offers on e-platforms?


Assemble the right team

 It can be easy to forget that humans will continue to play a central role in companies’ future success.

The “business first” approach requires assembling a team capable of close collaboration, leaps of imagination and that has a readiness to experiment.

In my view the data “dream team” should consist of a statistician, some data visualization and machine-learning/deep-learning engineers, and one or more expert from the business domain.

The leader of such a team should have enough knowledge in all these areas, but what is more critical is their ability to amalgamate these skill sets to form both hypotheses and solve business problems.

They should also be able to communicate the new data-driven culture across the whole organization, all the way up to the boardroom.


Be patient

 There are times when even the best-laid plans fall flat. The best remedy for this is to embrace a culture of “deploy first, measure effectiveness, and continually improve”.

If in doubt, remember that machine-learning systems will get better and better over time.

Finally, we must not forget to celebrate the role of human intuition and creativity in imagining and overcoming business problems, something machines will take a long time to learn, if they ever do.


Dr. Gautam Shroff, VP and Chief Scientist at Tata Consultancy Services Research, will be speaking at the TCS Slush Experience 2017 in Helsinki on 29 November